nanosecond switching behavior of nanopillar spin valves · nanosecond switching behavior of...

Cornell NanoScale Facility, Member: National Nanotechnology Infrastructure Network page 238

Physics & Nano- Structure Physics

Abstract:We have used nanopillar spin valves to study the reversal of the magnetization

of a thin ferromagnetic layer using short (1-100 ns) current pulses. Patterning of high aspect ratio (~3:1) ellipses has produced thermally stable devices in which reversal occurs between 0.6 mA for a 10 ns pulse and 2 mA for a 1 ns pulse. These results are approximately 2-3x smaller than previously reported [1, 2] and show that spin transfer can be a viable mechanism for writing the state of a bit in magnetic random access memory (MRAM).

Summary:The development of MRAM has provided a fast, non-volatile alternative

to silicon based memory technologies. However, current methods for writing bits require magnetic fields generated by large (~10 mA) currents, leading to issues with power dissipation. In addition, bit density will be limited due to issues with localization of the writing field. An alternative writing mechanism would utilize the spin transfer effect [3, 4], which uses spin-polarized currents localized to individual bits. Current switched MRAM would require writing currents (Ic’s) on the order of 0.1-0.2 mA to be compatible with CMOS technology, with the additional requirement that a bit maintain its state for > 10 years (barrier between states > 1 eV) to ensure non-volatility.

Our nanopillars consist of a 20 nm NiFe (Py)/12 nm Cu/4.5 nm NiFe trilayer. We utilize the fact that Ic scales with the saturation magnetization (Ms) squared, while the energy barrier is proportional to Ms, so a low moment magnet such as Py (Ms ~ 560 emu/cm3) should lower Ic without significantly reducing thermal stability. Additionally, we can boost the energy barrier by increasing the aspect ratio of our devices, which increases shape anisotropy energy. These devices are patterned using e-beam lithography, where we expose specially designed shapes to exploit electron proximity effects.

Nanosecond Switching Behavior of Nanopillar Spin ValvesCNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman, Daniel C. Ralph

Using this technique, we are able to pattern approximately elliptical shapes with minor axes ~ 40 nm and 3:1 aspect ratios. Ion milling and photolithography further defines the pillar and leads.

Typical energy barriers of these devices are ~ 0.8 eV as determined from a Kurkijarvi approach [5, 6]. To observe the switching behavior on ns time scales, we repeatedly pulse a device and measure switching probabilities. For a 10 ns pulse, we get 100% switching at 0.6 mA, increasing to 2 mA for a 1 ns pulse. This large increase in Ic is due to a change in the switching mechanism from thermally assisted switching to spin torque driven reversal around 1 ns. In this region, the magnetization of the thin ferromagnet must be overdriven to switch its orientation.

Further optimization of these structures requires increasing the spin torque incident upon the switching layer. An enhancement in spin torque may occur by introducing a misalignment angle between the magnetizations of the two ferromagnets. Adding a third magnetic layer has also been shown to reduce Ic [7]. In addition, further materials exploration and study of the effects of the electrodes on Ic should give further reduction of current amplitudes necessary for ns switching. With these advances, spin transfer switched MRAM is an attractive option for further generations of memory devices.

References:[1] S. Kaka, et al. J. Magn. Magn. Mater. 286, 375 (2005).[2] T. Devolder, et al. J. Magn. Magn. Mater. 286, 77 (2005).[3] J. A. Katine, et al. Phys. Rev. Lett. 84, 3149 (2000).[4] F. J. Albert, et al. Phys. Rev. Lett. 89, 226802 (2002).[5] J. Kurkijarvi, Phys. Rev B. 6, 832 (1972).[6] E. B. Myers, et al. Phys. Rev. Lett. 89, 196801 (2002).[7] G. D. Fuchs, et al. Appl. Phys. Lett. 86, 152509 (2005).

Cornell NanoScale Facility Physics & Nanostructure Physics, page 239 National Nanotechnology Infrastructure Network

• Writing bits in MRAM with electrical current eliminates power dissipation and bit density problems arising with magnetic field writing.

• Devices take advantage of small (40 x 120 nm) shapes possible with e-beam lithography to decrease currents for switching the magnetization of a thin nanomagnet from both parallel to antiparallel (P-AP) and antiparallel to parallel (AP-P) with respect to a thicker fixed orientation nanomagnet.

• Switching currents similar in both directions, which is advantageous for device applications.

Nanosecond Switching Behavior of Nanopillar Spin ValvesCNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman, Daniel C. Ralph

User(s):Patrick BragancaIlya KrivorotovOzhan OzatayAndrei GarciaNathan EmleyJack Sankey

Principal Investigator(s):Robert A. BuhrmanDaniel C. Ralph

Affiliation(s):Applied and Engineering Physics, Cornell University

Primary Funding:ARO through DARPA, and the NSF/NSEC program through the Cornell Center for Nanoscale Systems

Contact Information:[email protected]@[email protected]

Figure 2: Switching probability vs. pulse amplitude for ap-p Figure 3: Switching probability vs. pulse amplitude for p-ap

Figure 1: Top view (SEM) & cross-section of spin valve.



Abstract:We perform time-resolved measurements of current-induced reversal

of a free magnetic layer in Py/Cu/Py (Py = Ni80Fe20) nanopillars at bath temperatures T ranging from 4.2K to 300K. Switching speeds show a T-dependence that can be understood as coming from the T-dependence of the free layer magnetization. When the free layer moment switches faster than a critical switching time (~1 nsec), the distributions in switching times are both minimal (~0.5 nsec) and completely T-independent, indicating that thermal fluctuations do not significantly affect the switching dynamics. It is this critical switching time, rather than a critical current density, which characterizes the onset of this deterministic switching regime since it is consistent for all T and is reproducible from sample to sample.

Summary:A spin-polarized current passed through a nanoscale magnetic layer has been

shown to excite the magnetic moment into persistent dynamical states [1] as well as induce switching of the magnetization direction via some transient dynamical trajectory [2]. The dominant theory describes the mechanism for such excitations of the magnetic moment as coming from a spin torque applied by the spin polarization of the current [3]. Motivated by the potential for technological applications of spin torque switching as a writing scheme for magnetically-based non-volatile memory and programmable logic devices, the goal of the current work is to understand the role of temperature T in the magnetization dynamics during its reversal trajectory. We perform time-resolved measurements of the switching event of the thin magnetic layer in Py(thick)/Cu/Py(thin) nanopillar structures with lateral dimensions of ~100nm, while controlling T between 4.2K and 300K.

Time-resolved measurements of the switching event are performed by averaging over 10,000 individual traces due to signal to noise considerations. Consequently, the effects of T are implicitly incorporated into the averaged data since there is a certain amount of thermal randomization of the reversal trajectory from one switching event to another. The measured switching events

The Effects of Temperature and Current Amplitude on Spin-Torque Driven Magnetization Reversal

CNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman

are thus not abrupt (~few tens of psec) as predicted from simulations, but rather show a much more gradual transition of a few nsec. The width of this transition, or dispersion in switching times, is a measure of the randomizing effects of temperature. When plotting dispersion as a function of I, we find that the switching behavior naturally sub-divides into two regions. At lower currents, the dispersion values are large and there is some small T-dependence. For higher currents, there is no T-dependence and the dispersion is minimal. This T-independent region at larger I is advantageous for spin torque writing because it means that the switching time will be consistent for each actuation of the device.

The 50% drop in signal is defined as the time for the magnet to reverse direction (tswitch). When plotted versus current amplitude I, an inverse proportionality is found, tswitch

-1 = K(I – Ic), in agreement with predictions [4] and previous reported results [2, 5]. By accounting for the T-dependence of parameters K and Ic [5], we find that the majority of the T-dependence of tswitch can be understood as a rescaling of these parameters with the T-dependent free layer magnetization M(T). By comparing the dispersion with tswitch data at various T, we find that the onset of the regime showing T-independent and minimal dispersion is not defined by a critical current density, but rather by a critical switching time.

References:[1] S. I. Kiselev, J. C. Sankey, I. N. Krivorotov, N. C. Emley, R. A. Buhrman, & D. C. Ralph,

“Microwave oscillations of a nanomagnet driven by a spin-polarized current,” Nature 425, 380 (2003).

[2] I. N. Krivorotov, N. C. Emley, J. C. Sankey, S. I. Kiselev, D. C. Ralph, & R. A. Buhrman, “Time-Domain Measurements of Nanomagnet Dynamics Driven by Spin-Transfer Torques,” Science 307, 228-231 (2005).

[3] J. Slonczewski, “Current-driven excitation of magnetic multilayers”, J. Magn. Magn. Mater. vol. 159, L1 (1996).

[4] J. Z. Sun, “Spin-current interaction with a monodomain magnetic body: A model study”, Phys. Rev. B vol. 62, 570 (2000).

[5] R. H. Koch, J. A. Katine, & J. Z. Sun, “Time-Resolved Reversal of Spin-Transfer in a Nanomagnet”, Phys. Rev. Lett. vol. 92, 088302 (2004).


• Time-resolved measurements of current (I) induced reversal of the thin Py layer magnetic moment in Py(20nm)/Cu(10m)/Py(2nm) (Py = Ni80Fe20) nanopillars are performed at various temperatures T between 4.2K and 300K.

• Switching times of a few nanoseconds are achieved for I ~ 1 to 2 mA.

• The measured switching events are averages of several thousand traces.

• Thermal randomization of trajectories leads to dispersions in switching times.

• The dispersion minimal and T-independent at larger currents.

The Effects of Temperature and Current Amplitude on Spin-Torque Driven Magnetization Reversal


User(s):Nathan C. EmleyIlya N. KrivorotovJack C. SankeyPatrick M. BragancaAndrei G. F. GarciaOzhan Ozatay

Principal Investigator(s):Robert A. Buhrman


Primary Funding:Cornell Nanoscale Systems Center, NSF and ARO MURI

Contact Information:[email protected]



Abstract:We have fabricated nanoscale magnetic tunnel junctions (MTJs) with

an additional fixed magnetic layer added above the free layer of a standard MTJ structure. This acts as a second source of spin-polarized electrons that, depending on the relative alignment of the two fixed layers, either augments or diminishes the net spin torque exerted on the free layer.

Summary:Within the past few years, spin-polarized current has been used to

hysteretically switch the magnetization of a nanomagnet, first in spin valve [1] and then MTJ [2, 3] structures by the transfer of spin from conduction electrons [4, 5]. This opens the possibility of magnetic memory that can be directly written with spin-polarized current, which provides a scalable means to deliver non-volatile, low power, and high-speed memory. MTJs are an appropriate impedance match to CMOS technology, but operate at larger voltages than all-metallic spin valve devices. It would, therefore, be an advantage to boost the magnitude of the spin torque exerted on the nanomagnet to lower the operating voltage.

The multilayers for our samples consist of (in nm): CoFe 8/AlOx~0.7-0.8/Py 4/Cu 6/CoFe 5, where Py is Ni81Fe19 and CoFe is Co86Fe14. The films were patterned by a combination of electron-beam lithography and ion milling into elliptically shaped (~35 x 90 nm) nanopillars, where the current flows perpendicular to the plane of the multilayer films. The 4 nm of Py above the AlOx serves as the magnetic free layer, and the two CoFe layers are both fixed magnetic layers. The resistance state reflects whether the bottom fixed layer (8 nm CoFe) and the free layer are aligned parallel (P, low resistance) or antiparallel (AP, high resistance). The role of the top fixed layer (5 nm CoFe) is to provide an additional source of spin-polarized electrons to the free layer. When the two fixed layers are anti-aligned with respect to one another, the spin currents from each of the two fixed layers exert torque in the same direction, resulting in a lower critical current. When the fixed layers are aligned, the

Adjustable Spin Torque in Magnetic Tunnel Junctions with Two Fixed Layers

CNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman, Daniel C. Ralph

spin-polarized electrons from the two fixed layers exert torque in opposite directions, which nearly cancels the net spin torque.

After setting the alignment of the two fixed magnetic layers, we characterize the free layer nanomagnet by measuring resistance while sweeping magnetic field (H) at fixed current bias (I) and sweeping I at fixed H. Measurements are made at T = 77K since the small free layer is not thermally stable at room temperature.

From these data, we construct a free layer phase diagram for the two magnetic alignments of the fixed layers by the location of switching events in H and I. By fitting the phase boundaries to a thermally activated switching model [6], including the effects of joule heating, we can determine both the increase in temperature above the bath temperature as well as the T = 0K critical current for switching for each magnetic configuration of the fixed layers. When the fixed layers are aligned, the current densities normalized by free layer thickness are (Jc,o/t) P-AP = (-2.8 ± 1.0) x 107 A/cm2 nm and (Jc,o/t)AP-P = (1.4 ± 0.2) x 107 A/cm2 nm.

By comparison to a spin valve with a Py free layer [7], we conclude the cancellation of the torques from each of the two fixed layers reduces the net spin torque by a factor of ~4. For the anti-aligned magnetic configuration, (Jc,o/t) P-AP = (2.9 ± 0.6) x 106 A/cm2 nm and (Jc,o/t)AP-P = (-2.8 ± 0.6) x 106 A/cm2 nm. This represents an increase of the spin torque by a factor of ~2 relative to the single fixed layer case [7].

References:[1] J. A. Katine et al, Phys Rev. Lett. 84, 3149 (2000).[2] Y. Haui et al, Appl. Phys. Lett. 84, 3118 (2004).[3] G. D. Fuchs et al, Appl. Phys. Lett. 85, 1205 (2004).[4] J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996).[5] J. C. Slonczewski, Phys Rev. B 71, 024411 (2005).[6] G. D. Fuchs et al, Appl. Phys Lett. 86 152509 (2005).[7] I. N. Krivorotov et al, Phys Rev. Lett. 93, 166603 (2004).


Adjustable Spin Torque in Magnetic Tunnel Junctions with Two Fixed Layers

CNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman, Daniel C. Ralph

User(s):Gregory D. Fuchs

Principal Investigator(s):Robert A. BuhrmanDaniel C. Ralph


Primary Funding:ARO through DARPA


Figure 1, opposite: (a) When the two fixed layers are anti-aligned, the spin current from each layer exerts torque in the same direction. (b) When the two fixed layers are aligned, the spin torques nearly cancel.

Figure 2, below:Data is acquired by, (a) sweeping magnetic field (H) at constant current (I), and (b) I at H.Switching points are plotted in H and I to construct a phase diagram, both for (c) aligned fixed layers, and (d) anti-aligned fixed layers.By fitting to the phase boundaries (solid lines), we can determine T = 0K critical currents and also temperature increase due to joule heating.



Spin Transport and Spin Momentum Transfer in Current-Confined Nanopillar Spin-Valves


Abstract:Spin dependent transport across nano-scale magnetic devices leads not only

to very interesting new physics but also to potential applications in ultra-high density, non-volatile data storage applications. Current Perpendicular to Plane Giant Magnetoresistance (CPP-GMR) and current induced magnetization switching by transfer of spin angular momentum [1, 2] are two very important implications of the spin-transport at the nano-scale. Both of these effects can be effectively studied using a nanopillar or a nanocontact.

We have developed a novel nanocontact technique whereby a ~3 nm aluminum oxide barrier layer is inserted in the midst of the Cu spacer of a ~100 nm diameter CPP spin valve. This barrier layer is patterned with e-beam lithography followed by an in-situ ion milling and deposition process to form an intentional metallic pinhole, ~10-20 nm in diameter. This design allows us to explore the consequences of non-uniform current injection into a nanopillar via a nanoconstriction and also concentrates the current flow away from the device edges, increases the device impedance, and decreases the current required for spin transfer switching of a nanomagnet.

Summary:In a magnetic multilayer, the relative magnetic orientation of the constituent

layers affects the resistance of the device. This configuration can be manipulated either by applying an external field (GMR effect) or via spin momentum transfer from a spin-polarized current (Spin-Transfer effect).There has been a lot of experimental effort in studying both effects in metallic nanocontacts [3], nanopillars [4], and nanowires [5] motivated by potential technological applications such as Magnetic Random Access Memory (MRAM). It is desirable to have a high GMR signal and low critical currents (~106 A/cm2) for applications. The idea of using a nanocontact to inject spin-polarized current into a nanomagnet is a good candidate to achieve these goals.

We have developed a novel point contact technique which allows us to probe local spin-dependent transport in a spin-valve nanopillar as well as to explore the effect of nonuniform current injection on the spin torque. We use an ion-beam deposited insulating barrier (AlOx) on which a nanohole is exposed by electron beam lithography. An in-situ ion milling and deposition process results in a magnetic multilayer with around a 10-20 nm nanocontact. This is followed by a standard aligned nanopillar process where we use ion-milling and PECVD oxide to define and isolate the devices. The electrical contact to the devices is achieved by a top lead deposition and lift-off process.

The devices we have studied include a Py/Cu/Py spin valve with 20, 8 and 5 nm thicknesses respectively. These devices have been patterned to be 150 by 250 nm ellipses with around 30 nm nanoholes located in between the the copper spacer and thin permalloy layer. As expected, the resistance of such a device is about a factor of 3 larger than that of a nanopillar without the nanohole. The field scans show a clear magnetoresistance signal with a maximum and minimum corresponding to antiparallel and parallel alignments of the magnetic layers respectively.

Liquid helium temperature measurements of current scans show magnetic switching at around 100 microamps which corresponds to a critical current density of the order of 4 x 105A/cm2, two orders of magnitude smaller than that of a similar sized nanopillar without the hole. We attribute this result to domain nucleation and domain wall drag by spin transfer torque.

References:[1] J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996).[2] L. Berger, Phys. Rev. B 54, 9353 (1996).[3] E. B. Myers et al., J. Appl. Phys. 87, 5502 (2000).[4] J. A. Katine et al., Phys. Rev. Lett. 84, 3149 (2000).[5] J.E. Wegrowe et al, J. Appl. Phys. 96, 4490 (2004).


• A new point contact nanopillar technique has been developed which enables nonuniform current injection into a nanomagnet.

• This design has potential advantages such as enhanced impedance level, reduced spin transfer threshold currents for magnetic memory applications.

• Field scans show clear magnetoresistance signal. Spin transfer measurements indeed show that the critical currents for current induced switching are reduced compared to a pillar without the nanohole.

Spin Transport and Spin Momentum Transfer in Current-Confined Nanopillar Spin-Valves


User(s):Ozhan OzatayNathan EmleyPatrick BragancaGregory Fuchs



Primary Funding:NSF, DARPA

Contact Information:[email protected]@cornell.edu

Figure 1: Cross-sectional schematic of a current-confined nanopillar.

Figure 2: Magnetic field scan of a current-confined pillar.

Figure 3: Current scan of a current-confined pillar of 4.2K.



Abstract:Terbium doping in Permalloy spin valves dramatically increases the critical

currents needed for switching and dynamic behavior onset while reducing the microwave noise, without significantly affecting other properties.

Summary:Recent experiments on Co(fixed)/Cu/Co(free) magnetic nanopillars [1, 2]

have shown that a spin-polarized current (I) applied perpendicular-to-plane (CPP) can change the relative alignment of the two magnetic layers. The prevailing theories [3, 4] indicate a transfer of spin angular momentum at the Normal Metal/Ferromagnet interface from the conduction electrons to the magnetic moments, giving rise to a mutual torque. Upon reaching a critical current, the torque becomes great enough to overcome the intrinsic Gilbert damping and flip the thinner layer ferromagnet (free) either parallel (p) or anti-parallel (ap) to the thicker layer ferromagnet (fixed), depending on the direction of the current. The so-called ‘spin transfer’ switching has stimulated interest as a potential write mechanism for non-volatile memory elements, although it would act as a parasitic effect for CPP field sensors such as those used in modern hard drive read heads. Further work has shown that high frequency dynamics can also be excited by the spin-torque [3].

While magnetic damping is understood to play a fundamental role in spin-torque phenomena, little experimental work has been done to study the effect of varying the damping parameter a.

Recently, light terbium (Tb) doping in thin films of permalloy (Py) has been shown to increase a by several orders of magnitude [4]. To directly study the effect of increased a on spin-transfer systems, we have fabricated

Enhanced Magnetic Damping in Spin-Transfer ExcitationCNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman

0.05 µm2 Py/Cu/Py nanopillar spin valves with Tb-doping between 0 and 2% in the free layer. Electron beam lithography and ion beam etching techniques were used to define the nanopillars, while a number of oxide deposition methods were used to electrically isolate the top and bottom leads of the device.

We find that, while the Gigantic Magneto-Resistance (GMR) varies less than 20%, the critical currents for reversibly switching the free layer (proportional to a) are two to three times larger in the 2% Tb samples than in pure Py samples. This substantial increase is still considerably less than the increase in a observed in the bulk film measurements of similar composition samples, suggesting that processes other than intrinsic spin-orbital coupling can dominate a in spin-transfer nanopillars. The Tb doping also increases the critical current for the onset of processional dynamics. Spin-torque induced telegraph switching microwave noise, believed to be a significant contribution to read-head noise, is also dramatically reduced by Tb doping.

These results suggest Tb doping as one approach to reducing the negative impact of spin torque effects on nanoscale spin valve and tunnel junction read head sensors.

References:[1] J. A. Katine et al., Phys. Rev. Lett. 84, 3149 (2000).[2] F. J. Albert et al., Appl. Phys. Lett. 77, 3809 (2000).[3] S. I. Kiselev et al., Nature 425, 380 (2003).[4] S. E. Russek, et al., J. Appl. Phys. 91, 8659 (2002).


• 0.05 µm2 Py/Cu/Py nanopillar spin valves fabricated with Tb-doping between 0 and 2% in the free layer to studying increased damping in spin-transfer systems.

• The Tb doping increases the critical current for switching, for the onset of processional dynamics, and reduces the microwave noise power.

• The effects will be beneficial for next generation CPP hard drive read heads.

Enhanced Magnetic Damping in Spin-Transfer ExcitationCNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman

User(s):Eric M. Ryan



Primary Funding:A.R.O. M.U.R.I. grant

Contact Information:[email protected]@aol.com

Figure 1, top right: Cross-sectional schematic and SEM image of a nanopillar.

Figure 2, right: Tb-doped samples require higher currents to switch.

Figure 3, far right: Tb-doped samples have less microwave noise at a given current.

Figure 1



Abstract:We have fabricated micron-sized tunnel junctions incorporating aluminium

oxide barriers treated with a low energy electron bombardment. These junctions show a significantly lower 10 Hz 1/f resistance noise level at 4.2K than junctions which have not been treated. The electron bombardment process thus provides a means for producing low noise, aluminium oxide based Josephson junctions for quantum computing applications.

Summary:Micron-sized Al/AlOx/Al tunnel junctions are fabricated using standard

photolithography, reactive ion etching and ion milling techniques. Junction layers are formed by first thermally depositing an aluminium base electrode layer, followed by room temperature thermal oxidation to form the oxide barrier layer and finally the aluminium top electrode layer. The thickness of the oxide barrier is about 15-20Å. For a subset of junctions, the oxide layer is exposed to low energy (~10-20 eV) electrons prior to the deposition of the top electrode layer. X-ray photoelectron spectroscopy studies show that although the thermally oxidised barrier layer is oxygen deficient, the low energy electron bombardment can drive surface chemisorbed oxygen molecules into the oxide, making it more stoichiometric [1]. This results in an oxide barrier layer with fewer defect states, which have been demonstrated to be a source of 1/f resistance noise in Josephson junctions [2].

Fabrication of Low Noise Aluminium Oxide Tunnel JunctionsCNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman

The 1/f resistance noise characteristics of the junctions are measured between 4.2K and 300K. At 4.2K, junctions exposed to the electron bombardment exhibit a lower 10 Hz resistance noise level (~1.2 x 10-22 m2/Hz) compared to junctions which were not exposed (~1.3 x 10-21 m2/Hz). The improvement in the noise figure by a factor of ~10 indicates that the low energy electron bombardment method can be used effectively to produce lower noise Josephson junctions based on aluminium oxide tunnel barriers.

References:[1] E. Tan, P.G. Mather, A.C. Perrella, J.C. Read and R.A. Buhrman, Physical Review B 71,

161401(R) (2005)[2] C.T. Rogers and R. A. Buhrman, Physical Review Letters 53, 1272 (1984).


• We fabricated micron-sized aluminium oxide tunnel junctions and measured their low frequency noise character-istics between 4.2K and 300K.

• By exposing the aluminium oxide barrier to a low energy electron bombardment before junction fabrication, the low frequency resistance noise in the junction can be significantly reduced.

Figure 1, opposite: Aluminium oxide tunnel junction 1/f resistance noise figures at 10 Hz measured at temperatures between 4.2K and 300K. At 4.2K, junctions containing an electron bombarded oxide layer (filled circles) show a factor of ~10 reduction in the noise figure compared to junctions whose oxide layer had not been electron bombarded (hollow triangles).

Fabrication of Low Noise Aluminium Oxide Tunnel JunctionsCNF Project # 111-80 Principal Investigator(s): Robert A. Buhrman

User(s):Juting ZhaiEileen Tan



Primary Funding:Office of Naval Research, NSF (CCMR)




Abstract:We have developed a fabrication technique to form contacts to the ends of

whisker-like single crystals of the highly anisotropic, quasi-1D charge density wave (CDW) conductor NbSe3. The technique combines electroplating and standard thin film processing (lithography, etching), and addresses a variety of problems with standard side/bottom/top contacts in evaluating collective transport properties. This technique maybe useful in fabricating controllable mesoscale semiconductor junctions and contacts to nanowires of various materials.

Despite the wide variation in electronic anisotropies of solid state materials, transport measurements are typically performed with contacts applied to the top or bottom of a crystalline sample [1]. Quasi-one dimensional conductors have a series of remarkable electronic properties, including collective charge transport by sliding charge density waves [2]. Most properties (resistivity, optical conductivity) of these long, flat, ribbon-like crystals show a significant degree of anisotropy. Given that as-grown crystal ends are always highly irregular, we set out to develop a method that would yield uniform current injection along the highly conducting one-dimensional b* axis that minimized contact resistances.

We have used standard thin film processing techniques to remove photoresist, polyimide (used to affix samples to the substrate) and parts of the sample. A 1500Å gold seed layer is deposited using e-beam evaporation in the “pits”

Electroplated End-Injection Current Contacts to the Quasi-1D Charge Density Wave Conductor NbSe3

CNF Project # 316-87 Principal Investigator(s): Prof. Robert E. Thorne

thus produced. Electroplating is then used to deposit a thick (~ 1 µm) layer of gold in the a-c* plane. Our measurements show that NbSe3 samples contacted in such a way show decreased depinning thresholds, and quantitative changes in other transport parameters, especially those associated with the phase slip process required for current conversion [3]. These contacts will allow the first meaningful measurements of collective transport in very short crystals and in fully gapped materials like orthorhombic TaS3 whose anisotropy diverges at low temperatures.

Summary:A combination of electroplating and standard lithographic techniques is

used to fabricate end contacts to the quasi-1D conductor NbSe3. Contacts are controllable in thickness, length and width, and do not alter fundamental nature of CDW transport. This technique should be especially valuable for fully gapped quasi-1D materials, as well as for other highly anisotropic conductors.

References:[1] Nanoelectronics and Information Technology, ed. by Rainer Waser, Wiley, VCH, 2003.[2] R. E. Thorne, Charge density wave conductors, Physics Today, May 1996, 42-49.[3] A. F. Isakovic, K. Cicak, and R. E. Thorne, End Current Injection Contacts to quasi-1D

CDW Conductor NbSe3: Fabrication, Characterization and Control of Transport Properties, to be submitted to Phys. Rev. B.


• In traditional side contacts, spreading resistances and inhomogeneous current injection can dominate measured electronic properties, particularly those associated with the collective transport mode.

• End contacts, which inject current along the highly-conducting axis, eliminate these nonuniformities and should allow the “intrinsic” collective response to be measured.

Electroplated End-Injection Current Contacts to the Quasi-1D Charge Density Wave Conductor NbSe3

CNF Project # 316-87 Principal Investigator(s): Prof. Robert E. Thorne

User(s):Abdel IsakovicKatarina Cicak

Principal Investigator(s):Prof. Robert E. Thorne

Affiliation(s):Physics Dept., Cornell University

Primary Funding:NSF DMR




Abstract:We are developing tools for manipulating and mounting crystals of proteins

and other biological macromolecules for structure determination by X-ray crystallography. Some of these tools have been commercialized by Mitegen, LLC, and are now used by academic and industrial researchers in more than 25 countries.

The molecular structures of proteins, viruses and macromolecular complexes are essential to understanding biological function and to the design of new medications to modify those functions. With advances in genetic cloning, expression, and purification on the one hand and in x-ray sources, detectors and analysis software on the other, the difficulties associated with growing and mounting crystals of proteins for structure determination by X-ray crystallography have become important obstacles limiting the rate at which new structures are solved. Existing tools for manipulating crystals after they are grown, and mounting them in cold gas streams for X-ray measurements, degrade achievable data quality and limit overall diffraction throughput. We are developing a family of tools for retrieving crystals from crystallization media, performing post-growth manipulations like cryoprotectant and heavy atom soaks, and holding the crystals in the X-ray beam [1-3]. Some of these tools have been commercialized by Mitegen, LLC of Ithaca (www.mitegen.com).

The tools are fabricated using standard photolithographic techniques from Photoneece PWDC-1000 photoexposable polyimide, which provides excellent mechanical properties and excellent X-ray transparency. Protein crystals can

Microfabricated Tools for Macromolecular CrystallographyCNF Project # 316-87 Principal Investigator(s): Prof. Robert E. Thorne

be more than 90% water and are spectacularly fragile, and so the tools must facilitate very gentle handling. We have designed and prototyped more than 50 tools for retrieving small and large crystals from crystallization drops, removing crystals that have adhered to the surfaces of crystallization media, separating crystal clusters, and mounting for flash cooling and data collection crystals with a variety of morphologies and sizes ranging from microns to millimeters.

Summary:

New tools for manipulating and mounting crystals of biological macromolecules for structure determination by X-ray crystallography are being developed. Some of these tools have been commercialized, and are rapidly being adopted by the worldwide structural genomics community.

References:[1] Microfabricated Mounts for Microcrystal Cryocrystallography, R. E. Thorne, Z. Stum, J.

Kmetko and K. O’Neill, J. Appl. Cryst. 36, 1455 (2003).[2] A New Crystal Mounting Method for Macromolecular Cryocrystallography, Z. Stum, J.

Kmetko, K. O’Neill, R. Gillilan, A Bartnik and R. E. Thorne, Synchrotron Rad. News 17, 31 (2004).

[3] A New Sample Mounting Technique for Room-Temperature Macromolecular Crystallography, Y. Kalinin, J. Kmetko, A. Bartnik, A. Stewart, R.Gillilan, E. Lobkovsky and R. E. Thorne, J. Appl. Cryst. 38, 333 (2005).


Figure 1:Tool for mounting roughly 200 µm protein crystals. The crystal rests in the small aperture in the tip. The channel and larger aperture allow excess liquid, which contributes to background X-ray scatter and causes crystal damage during flash cooling, to be wicked away.

Microfabricated Tools for Macromolecular CrystallographyCNF Project # 316-87 Principal Investigator(s): Prof. Robert E. Thorne

User(s):Guanhan Chew

Principal Investigator(s):Prof. Robert E. Thorne

Affiliation(s):Physics Dept., Cornell University

Primary Funding:NIH GMS


Figure 2:Tool for retrieving multiple small crystals from early crystallization trials. By using a small (0.1 mm) X-ray beam, the diffraction properties of crystals can be individually interrogated.



Abstract: We fabricate ferromagnetic nanowires with constrictions such that the

cross-section at the constriction can be controllably reduced from 100 nm to the atomic scale by means of electromigration. This allows measurements of the resistance of a domain wall trapped at the constriction as a function of the cross-section’s size. We observe magnetoresistance (MR) of around 0.3% for a (100 x 30) nm2 wide constriction and around 10% for an atomic-sized constriction. No MR higher than 15% is observed. These results are consistent with a geometrically-constrained domain wall trapped at the constriction [4].

Summary:The micromagnetic structure of materials on a nanometer scale differs

from that in the bulk. One example is the width of a domain wall located at a constriction separating two regions of wider cross-section. In bulk the domain wall’s structure is determined by a competition between exchange and anisotropy energies and does not depend on the geometry of the system. However, the situation changes once the cross-section of the constriction is just tens of square nanometers. The width of such a “constrained” domain wall is determined by the size of the constriction and can be of atomic dimension in the limit of a point contact [1]. Experimentally, formation of such domain walls can be monitored through resistance measurements. When the size of the magnetic wall gets smaller than the Fermi length, electrons cannot propagate adiabatically from one side of the wall to the other, resulting in a higher resistance [2]. Recently, several groups reported MR in excess of 100% in such atomic-sized constrictions [5], while others failed to reproduce these findings [6].

We developed a technique based on controlled electromigration [3] to tune the size of the constriction between two single-domain magnets from 3000 nm2 down to nanometer dimensions, and to measure its MR as a function of the

Magnetoresistance in Ferromagnet BreakjunctionsCNF Project # 598-96 Principal Investigator(s): Daniel C. Ralph

constriction size. The cross-section of the constriction determines its resistance which varies from 100 Ω into the kΩ range for atomic point contacts.

We fabricate two permalloy magnets separated by a narrow bridge using electron-beam lithography. The shapes of the magnets are chosen to promote shape anisotropy and favor formation of the domain wall in the area of the constriction at intermediate magnetic fields. The magnets are electrically contacted with gold leads using a separate step of e-beam lithography. DC conductance measurements are performed in a cryostat at 4.2 K.

The voltage across the constriction is slowly ramped up while monitoring the current. At roughly 3 mA, electromigration sets in (as indicated by a slower increase in current) and the bias voltage is quickly lowered by acquisition software. Repeating this procedure allows us to lower the resistance of the junction to any value between 100 Ω and 10 kΩ with 10% accuracy.

The MR of such a device strongly depends on its resistance. For resistances below 100 Ω the magnetoresistance typically is around 0.3-0.6%. The value of MR increases when the size of the constrictions gets smaller. In addition, the sign of MR changes in some devices. Finally, we observe MR as high as 10% when the resistance of the device is comparable to the quantum of resistance. This observation is consistent with electrons scattering off a constrained domain wall trapped in the region of the constriction [4]. At this point, however, we cannot exclude other sources of MR, such as tunneling contributions in the high-resistance regime.

References:[1] P. Bruno Phys. Rev. Lett. 83, 2425 (1999).[2] G. Tatara et al. Phys. Rev. Lett. 83, 2030 (1999).[3] D. R. Strachan et al. Appl. Phys. Lett. 86, 043109 (2005).[4] R. F. Sabirianov et al. cond-mat/0506263 (2005).[5] Y.-W.Zhao et al. J. M. M. M. 223, 169 (2001).[6] M. I. Montero Phys. Rev. B 70, 184418 (2004).


Figure 1, top right: Two permalloy magnets separated by a narrow constriction.

Figure 2, below left: The cross-section of a constriction is reduced in stages (insets) by ramping the bias voltage until electromigration begins, and then quickly decreasing the bias.

Figure 3, below right: Magnetoresistance for 3 different samples as a function of resistance (A and B) and magnetic field (C).

Magnetoresistance in Ferromagnet BreakjunctionsCNF Project # 598-96 Principal Investigator(s): Daniel C. Ralph

User(s):Kirill BolotinFerdinand Kuemmeth

Principal Investigator(s):Daniel C. Ralph

Affiliation(s):Department of Physics, Cornell University

Primary Funding:NSF/DMR and ARO


Figure 1

Figure 2 Figure 3



Abstract:Mechanically-adjustable breakjunctions [1] and electromigration-induced

breakjunctions with an electrostatic gate electrode [2] have both been used previously to make electrical measurements on single molecules. We have fabricated molecular devices that combine both mechanical adjustability and electrostatic gating [3]. Our goal is to probe electron transport in single molecules as a function of controlled variations in both the molecular conformation and the energies of the electronic states in the molecule.

So far we have studied C60 molecules in which we have observed a mechanically controlled charging energy. These transistors are therefore controllable with both gate voltage and mechanical motion.

Summary:Mechanically-adjustable breakjunctions [4] are devices in which a narrow

bridge of metal is suspended above a flexible substrate. By bending the substrate, the metal bridge can be broken, and the distance between the ends can be controllably adjusted, with increments of much less than a picometer. In an electromigration breakjunction, a current is passed through a narrow metal wire until it fails and forms two closely spaced contacts [5]. By fabricating a gate electrode underneath the electromigration region, it is possible to make 3-terminal transistor devices, and by depositing molecules of interest on top of the wire before electromigration, it is possible to make contact to single molecules [2]. Our work combines the virtues of both techniques [3].

First we make metal bridges suspended only 40 nm above a conducting substrate that will serve as the gate electrode. By using electromigration to break the bridges, we can form molecular junctions without initially having to bend the substrate. Subsequently, small deformations of the substrate adjust the size of the gap between electrodes, while the junction remains close enough to the gate electrode to allow molecular energy levels to be shifted by an applied gate voltage. Electromigration is a key factor in allowing us to use a silicon substrate since silicon is brittle and would fail before the metallic constriction

Mechanically-Adjustable and Electrically-Gated Molecular TransistorsCNF Project # 598-96 Principal Investigator(s): Daniel C. Ralph

did if we tried to break the junction mechanically. The use of silicon as the substrate makes it possible for us to make the breakjunctions in close enough proximity to the substrate to allow gating. Our samples are fabricated on thin silicon wafers to allow some substrate bending. Silicon oxide is used as a spacing layer between the metal junction and the silicon substrate. Using a series of photolithography and e-beam lithography steps, and dry and wet etching, we define narrow gold bridges (50 nm wide) suspended 40 nm above the silicon substrate. We also make electrical contact to the wafer so that it can act as the gate electrode. Measurements are currently being made in a variable temperature insert (1.5 - 300 Kelvin) equipped with a superconducting magnet (14 Tesla). The mechanical breakjunctions are driven with a fine threaded screw actuated by a stepper motor via a reduction gear-box.

Measurements on individual C60 molecules show current-voltage curves with Coulomb-blockage features that vary systematically as the breakjunction contacts are moved apart. By acquiring gate-voltage scans of the devices at different source-drain displacements, we have observed how mechanical motion changes the conductances and capacitances of the tunnel junctions, and the molecule’s offset charge. Both of the tunneling junctions (source-molecule and molecule-drain) are modified by mechanical motion. We are now studying many other types of molecules exhibiting different transport mechanisms.

References:[1] M. A. Reed, C. Zhou, C. J. Muller, T. P. Burgin, J.M. Tour, Science 278, 252 (1997). [2] J. Park, A. N. Pasupathy, et al., Nature 713, 722 (2002).[3] A. R. Champagne, A. N. Pasupathy, and D. C. Ralph, Nano Letters 5, 305 (2005).[4] J. Moreland and P. K. Hansma, Rev. Sci. Instrum. 55, 399 (1984).[5] H. Park et al., Appl. Phys. Lett. 75, 301 (1999).


• We study electron transport in single molecules using mechanically-adjustable breakjunctions that allow electrical gating.

Mechanically-Adjustable and Electrically-Gated Molecular TransistorsCNF Project # 598-96 Principal Investigator(s): Daniel C. Ralph

User(s):Alexandre ChampagneJoshua Parks



Primary Funding:CNS/NSF/DMR


http://people.ccmr.cornell.edu/~ralph/projects/mcbj/

Figure 1: SEM image (78° tilt) of a Au bridge suspended 40nm above a Si substrate, which is used, after electromigration, as a mechanical breakjunction.

Figure 2: I-V curves for a C60 molecule in a Au breakjunction at different source-drain spacing.

Figure 3: dI/dV versus bias voltage and gate voltage for a C60 device at two different electrode spacings (the change in the threshold slope reflects a change in capacitance).



Abstract:We are studying the effect of light on single molecule transistors. These

three-terminal devices are fabricated using standard nanolithography techniques followed by electromigration. We have two experimental set-ups for illuminating the devices with UV/VIS/NIR radiation at cryogenic temperatures. Light-induced changes in conductance can be studied as a function of the gate and bias voltages.

Summary:We are extending the study of single molecule transistors [1, 2] to include

light-induced effects. Optically active molecules may be electronically excited, undergo a structural change, make and break chemical bonds, and/or exhibit photoconductivity. Such effects could show up as a change in conductance of the device, which can be studied as a function of bias voltage, gate voltage, and illumination wavelength. Because the nature of the response should be molecule-specific, light studies should also provide clear confirmation that the device contains the molecule of interest, rather than other species which may produce artifacts.

To make a single molecule transistor, a molecule must bridge a gap of comparable dimension between two conducting electrodes. Gaps of this size cannot be made with conventional lithography techniques, so we fabricate a continuous wire, then create the gap using electromigration [3]. The fabrication is done on an insulating SiO2 layer on Si. Thin Al gate electrodes are deposited where the devices will be, then a native oxide is allowed to form. Au or Pt wires (150 nm wide, 500 nm long, and 10 nm thick) are fabricated on top of the gate electrodes with e-beam lithography and liftoff. These wires are contacted by larger Au leads.

A chip with 30 devices is cleaned in an oxygen plasma, then molecules are deposited. The sample is loaded into the measurement apparatus which

Optical Studies of Single Molecule TransistorsCNF Project # 598-96 Principal Investigator(s): Daniel C. Ralph

is typically cooled to cryogenic temperatures to improve energy resolution. The wires are broken in situ by ramping the voltage bias until the wire fails, leaving a gap of order 1 nm between the resulting electrodes. In some of the devices, a molecule will bridge this gap. The presence of a molecule can be detected by Coulomb blockade, in which current turns on above a gate dependent threshold bias voltage.

We have two set-ups for measuring single molecule transistors under illumination. In the first, the devices are mounted at the bottom of a cryostat that is placed in a liquid helium storage dewar. An optical fiber is positioned to illuminate the devices with an external light source. The second set-up, currently being built, provides much higher photon intensities. The sample is placed close to a transparent window in a small cryostat which sits under a microscope. Laser light is focused onto the device of interest with a microscope objective. Light emitted from a device could also be detected in this set-up.

Thus far we have not seen reproducible changes in conductance of these devices with light. We have studied C60 and an organic complex with a Ru atom at the center. However, it is likely that the illumination intensities were not high enough, and several improvements are being made to overcome this issue, including the second set-up mentioned above. We also plan to optimize the electrode geometry to maximize the field enhancement at the gap between the antenna-like electrodes.

References:[1] H. Park et al., Nanomechanical oscillations in a single-C60 transistor. Nature 407, 57

(2000).[2] J. Park et al., Coulomb blockade and the Kondo effect in single-atom transistors. Nature

417, 722 (2002).[3] H. Park et al., Fabrication of metallic electrodes with nanometer separation by

electromigration. Applied Physics Letters 75, 301 (1999).


Purpose: Study conductance changes of single molecule transistors when exposed to light.

Figure 2, below left: Sample and optical fiber at base of cryostat.

Figure 3, below right: Coulomb blockage in a [Ru(tpy-py)2]2+ single molecule transistor (T = 6 K). I-V curves as a function of gate voltage from -455 mV to -415 mV in steps of 5 mV. We have not yet achieved reproducible changes with illumination.

Optical Studies of Single Molecule TransistorsCNF Project # 598-96 Principal Investigator(s): Daniel C. Ralph

User(s):Janice Wynn GuikemaJacob E. Grose



Primary Funding:NSF/NIRT & NSF/MRSEC through the Cornell Center for Materials Research


Figure 1, above: SEM images of (a) an unbroken wire and (b) a wire broken by electromigration. In some devices a molecule deposited on the surface bridges the gap, creating a single molecule transistor.



Abstract:The transfer of spin angular momentum from a DC spin-polarized current

passing through a nanoscale magnetic multilayer can drive spontaneous coherent magnetic oscillations. By measuring linewidths of spectra from the associated resistance oscillations, we demonstrate that the coherence is limited by thermal deflections of the magnetization about its equilibrium trajectory below 100 K, and by thermally-activated transitions between dynamical modes at higher temperature. The coherence time can be longer than predicted by macrospin simulations, suggesting that spatially non-uniform magnetic modes are relevant.

Summary:Spin-polarized current passing perpendicularly through magnetic multilayers

can exert torques on the magnetic moments through direct transfer of spin angular momentum at magnetic interfaces. This phenomenon allows small magnetic domains to be manipulated far more efficiently than is possible through the traditional methods involving applied magnetic fields. Furthermore, DC currents passing through multilayers can spontaneously excite new types of highly coherent precessional excitations not attainable with fields alone, enabling applications such as nanoscale microwave oscillators with frequencies that are tunable by current and/or magnetic field. The feasibility of such applications depends on achieving precession with a long coherence time, which is the subject of our investigation.

We study magnetic multilayers patterned into nanoscale pillars. We fabricate these “nanopillars” by first sputtering a multilayer of 80 nm Cu / 20 nm Py / 6-10 nm Cu / 2-7 nm Py / 2-20 nm Cu / 30 nm Pt and then evaporating 50 nm of carbon on top. Electron beam lithography is used to define an elliptically-shaped etch mask of dimensions 120 x 60 nm or 130 x 40 nm that is used for ion milling the surrounding multilayer. PECVD-grown SiO2 is then deposited to insulate the pillars before electrical contact to the top is finally made using photolithography.

Coherence of Microwave-Frequency Nanomagnetic Dynamics Driven by a DC Spin-Polarized Current

CNF Project # 598-96 Principal Investigator(s): Daniel C. Ralph, Robert A. Buhrman

Experiments have shown that DC current passing through such devices can excite steady-state dynamical modes [1,2], which produce peaks when the microwave power density versus frequency is measured with a spectrum analyzer. The narrowest linewidth (FWHM) of the peaks (related to the inverse coherence time) varies between samples, ranging from 550 MHz in room-temperature Co layers [1] down to 6 MHz in Py layers at 40 K [2]. By observing the dependence of the FWHM on temperature T and on the angle of the applied magnetic field relative to the magnetic easy axis, we are able to probe which mechanisms limit the linewidth [3]. Over the range of T studied (25-290 K, depending on the sample), the FWHM is found to increase by as much as a factor of 5. We argue that two mechanisms are important: (a) fluctuations of the moment about its equilibrium trajectory dominate below ~ 100 K and (b) thermally-activated transitions between different modes produce additional broadening at higher T. The low-T linewidths that we measure can be substantially narrower than predicted by simple models which assume that the magnet responds as a single macrospin, from which we conclude that spatially non-uniform modes are relevant and surprisingly effective in increasing the coherence time.

Further evidence for both non-uniform modes and mechanism (b) is observed in devices where, for certain ranges of current and field, two similar modes appear in the spectrum. When both modes are visible, their linewidths increase dramatically, which is likely due to high-frequency mode switching.

Linewidths are a also observed to be a strong function of the angle of the applied magnetic field relative to the easy axis of the nanomagnet. This effect is currently under investigation.

References:[1] S. I. Kiselev et al., Nature 425, 380 (2003).[2] I. N. Krivorotov et al., Science 307, 228 (2005).[3] J. C. Sankey et al., cond-mat/0505733 (2005).


• Motivation: Understand the coherence time for magnetic dynamical modes driven by a DC current.

• The linewidths increase with temperature T, dominated by magnetic deflections at low T and thermally-activated mode switching at high T.

• Switching between modes increases linewidths when two modes are present.

Figure 1, opposite: (a) Geometry of nanopillar. (b) T-dependence of linewidths in two devices.

Figure 2, below: Current dependence of (a) frequency and (b) linewidths for two modes in third device.

Coherence of Microwave-Frequency Nanomagnetic Dynamics Driven by a DC Spin-Polarized Current

CNF Project # 598-96 Principal Investigator(s): Daniel C. Ralph, Robert A. Buhrman

User(s):Jack Sankey1

Kiran Thadani1

Ilya Krivorotov2

Principal Investigator(s):Daniel C. Ralph1

Robert A. Buhrman2

Affiliation(s):1. Department of Physics, 2. Applied & Engineering Physics, Cornell University

Primary Funding:NSF/NSEC, DARPA, ARO

Contact Information:[email protected]@[email protected]@[email protected]



Abstract:Our goal is to fabricate devices that will make possible the measurement

of the high-frequency properties of carbon nanotubes. It has been predicted [1, 2] that carbon nanotubes should exhibit a resonant frequency that varies inversely with length and is in the range of hundreds of gigahertz for tubes a few microns long. This resonant frequency is important for two reasons. First, if it exists, it imposes design constraints on any nano-scale circuits that might include carbon nanotubes as either transistors or interconnects. Second, the frequency at which the resonance occurs is a probe of the interaction between electrons, which is different for one-dimensional systems such as nanotubes than for bulk materials with which we are familiar.

Because this frequency depends on the length of the nanotube and no technology currently exists to grow a large number of nanotubes of uniform length, it is necessary to study a single nanotube. The joint constraints of working in a very difficult-to-access frequency range and coupling radiation into a single nanotube require novel experimental methods to enable these measurements.

Devices for Measuring the High-Frequency Electrical Properties of Carbon Nanotubes

CNF Project # 689-97 Principal Investigator(s): Beth Parks

Summary:We have designed bowtie antennas bridged by a carbon nanotube that can

be used in a time-domain terahertz interferometer to couple radiation into the carbon nanotube. We have not yet been successful in fabricating an antenna in which a carbon nanotube bridges this gap between halves of the antenna. Fabrication is ongoing, and improvements to the method are in progress.

References:[1] “An RF circuit model for carbon nanotubes” P. J. Burke, IEEE Transactions on

Nanotechnology, vol.2, no.1, March 2003. p. 55-8.[2] “Luttinger liquid theory as a model of the gigahertz electrical properties of carbon

nanotubes” P. J. Burke, IEEE Transactions on Nanotechnology, vol.1, no.3, Sept. 2002. p. 129-44.


Devices for Measuring the High-Frequency Electrical Properties of Carbon Nanotubes

CNF Project # 689-97 Principal Investigator(s): Beth Parks

User(s):Anthony AnnunziataDobromir KamburovJeff SeelyBeth Parks

Principal Investigator(s):Beth Parks

Affiliation(s):Physics Department, Colgate University

Primary Funding:NSF NSEC (CNS)

Contact Information:[email protected]@[email protected]@mail.colgate.edu

http://departments.colgate.edu/physics/research/condensedm.html

Figure 1: In the fabrication of the devices, the first step is to create the small triangular catalyst pads and to grow nanotubes from them. Then an aluminum bowtie antenna is fabricated over them. The bowtie antenna has a very broadband response, so any features in the overall response will be due to the nanotube.

Figure 2: The catalyst pads are clearly visible in the image of the bowtie antennas.

Figure 3: Nanotubes are imaged using an atomic force microscope.

Figure 1

Figure 2 Figure 3



Background:Low device yield and placement issues are critical for the development of

carbon nanotube devices. Novel processes may enable the creation of new low cost micro and nano scale fluidic devices.

Abstract:This year we successfully patterned catalyst pads created by mixing ferric

nitrate into hydrogen silsesquioxane (HSQ) and patterning the film with low energy electron beams. The patterned catalyst was subsequently placed into an argon, ethylene, hydrogen mixture for CVD growth in a 750˚C furnace, to grow carbon nanotubes. The resulting carbon nanotubes were studied with AFM and FE-SEM.

Using this technique, tubes were easily suspended over grids patterned in the electrically insulating SiOx network that is created from the HSQ. This approach combines catalyst patterning, electrical isolation, and suspension supports in one lithographic step. This technique also was useful in observing the interactions of topography with the CVD of nanotubes, eliminating the wetting issues which dominated in our earlier studies.

In collaboration with both the Davis group from BYU and the McEuen group based at Cornell, we are working on new approaches for fabricating suspended nanotube oscillators, using wet etching chemistries in place of reactive ion etching. This approach will allow for nanotubes to be released and suspended microns above gating electrodes in hopes of dramatically increasing yield of these types of devices. Adhesion forces between nanotubes and SiO2 surfaces were quantitatively measured [1]. Nanotubes suspended on SiO2 terraces less than 250 nm wide were observed to slip with an axial tension of 8 nN. Even

Controlled Patterning of Nanotubes and Novel Processing with Hydrogen Silsesquioxane

CNF Project # 789-99 Principal Investigator(s): David M. Tanenbaum

tubes embedded in selectively deposited SiO2 films slid through the oxide with a tension of 10 nN. Understanding these adhesion forces is valuable for both single nanotube based electromechanical devices and composite materials reinforced with nanotubes.

In collaboration with the Craighead group, work has begun on fabrication of hybrid mechanical and fluidic devices. The focus had been on etch selectivity and integration techniques to enable useful measurements. Both wet and dry deep silicon etching has been used to fabricate membranes and through-wafer ports combined with both photolithographic and e beam patterning. An approach based on imprinting patterns into HSQ for conversion into fluidics has also been studied. Release of the imprint master was improved by coating the masters with (1H,1H,2H,2H-perfluorooctyl)trichlorosilane (or FOTS). The low cost of manufacturing such devices is very attractive, but HSQ imprinting proved to be unreliable, and it was difficult to bond over the imprinted patterns to seal the fluidics due to uneven surfaces after the imprinting step.

Summary:A wide variety of approaches are being used to lower the cost and increase

the yield of devices made with carbon nanotubes. Process development for hybrid mechanical and fluidic devices is in progress.

References:[1] J.D. Whittaker, et. al., “Measurement of the adhesion force between carbon nanotubes

and a silicon dioxide substrate”, in submission.


• We have developed a process that allows direct writing of catalysts for nanotubes in HSQ resist.

• Nanotubes suspended on SiO2 terraces are seen to slip with an axial tension of ~8 nN.

• We have imprinted patterns in unbaked HSQ films from e-beam patterned HSQ masters using FOTS as a release agent.

Controlled Patterning of Nanotubes and Novel Processing with Hydrogen Silsesquioxane

CNF Project # 789-99 Principal Investigator(s): David M. Tanenbaum

User(s):David M. Tanenbaum

Principal Investigator(s):David M. Tanenbaum

Affiliation(s):Department of Physics &Astronomy, Pomona College

Primary Funding:NSF

Contact Information:[email protected] Figure 1: Carbon nanotubes suspended above a grid patterned in HSQ. The tubes

were grown from ferric nitrate particles embedded in the e-beam patterned HSQ film. Charging is evident where tubes are in contact with the HSQ.

Figure 2: High-resolution nested lines written in HSQ resist by 100 keV electron beam lithography. Line edge roughness suggests the HSQ material is aggregating in solution.

Figure 3: An imprint lithography master fabricated by low energy electron beams in HSQ after use to imprint HSQ films. The master shows significant adhesion with the imprinted film. This was reduced dramatically when the masters were treated with FOTS.



Carbon NanoElectronicsCNF Project # 900-00 Principal Investigator(s): Paul L. McEuen

Abstract: Single walled carbon nanotubes (SWNT) which are rolled up graphene

sheets have been intensely studied in recent years due to their extraordinary electrical and mechanical properties. This recent study of SWNTs has brought renewed interest in the electronic properties of graphite, a low carrier density high purity semimetal consisting of graphene sheets separated by 0.3 nm and held together by weak van der Waals forces. Utilizing electron beam lithography and photolithography we have made electrical contact to both SWNTs and nanometer thin mesoscopic graphite to study the electronic transport properties of these carbon based systems.

The high frequency properties of SWNT transistors were examined up to 50 GHz and experiments were performed to examine the ultimate performance of nanotube transistors operating in the diffusive regime. We also used low temperature atomic force microscope (AFM) to probe single electron charging in carbon nanotube quantum dots. Our experiments on mescoscopic graphite involved Hall effect measurements and Coulomb blockade studies on quasi 2D graphite quantum dots.

Summary:Semiconducting carbon nanotubes have been shown to have very high

mobilities, high transconductances and long mean free paths. As a result, they offer promise as very high-frequency transistors. To explore the high-frequency properties of carbon nanotube transistors, we used a device layout in which the carbon nanotubes were gated with an Al top gate electrode underneath which lies a 10 nm evaporated silicon oxide gate insulator. A schematic of the device used in the experiments is shown in Figure 1. From these devices we demonstrated that SWNT transistors can operate as mixers at frequencies up to 50 GHz [1].

Further experiments on semiconducting SWNT field effect transistors examined the temperature and diameter dependence of the mobility and maximum conductance, and found these to be well described by predictions of acoustic phonon scattering in concert with the “relativistic” band structure of nanotubes. These results set the ultimate performance limits of SWNT

transistors operating in the diffusive regime [2]. For experiments on graphite quantum dots, we used electron beam

lithography to wire up mesoscopic graphite pieces dispersed on a degenerately doped Si wafer with a 200 nm thermally grown oxide. In this method, AFM is employed to locate thin pieces with respect to predefined alignment marks, and then electron beam lithography followed by metal evaporation is used to define multiple (two to six) electrodes on the piece (see Figure 2). The resulting quasi-2D graphite quantum dots have typical lateral dimensions of approximately 1 µm and vary in thickness from a few to tens of nm (see Figure 3). These devices were then measured in a field effect geometry at low temperatures (1.5 K and 100 mK), and carrier densities of 9.2-13 x 1012 cm-2 and mobilities of 200-1900 cm2/V-s were deduced.

In devices with tunnel contacts, Coulomb charging phenomena was observed and the gate and souce-drain capacitances were inferred [3]. We have also used this same lithographic technique to wire up SWNTs for low temperature scanning probe measurements. In these experiments, a low temperature AFM operating at 300 mK is used to sense single electron charging in carbon nanotube quantum dots.

The compatibility of graphene in its one and two dimensional forms with standard MOSFET technology available at CNF has allowed us to probe the electronic properties of this promising low dimensional material. Photolithography gives us the capability to easily produce large quantities of SWNT transistor devices while electron beam lithography allows us the flexibility to wire up not only SWNTs but quasi-2D graphite devices as well.

References: [1] “Mixing at 50GHz Using a Single-Walled Carbon Nanotube Transistor”; S. Rosenblatt,

H. Lin, V. Sazonova, S. Tiwari, and P.L. McEuen. (submitted)[2] “Band Structure Phonon Scattering and the Ultimate Performance of Single-Walled

Carbon Nanotube Transistors”; X. Zhou, J. Park, S. Huang, J. Liu, and P.L. McEuen. (submitted)

[3] “Coulomb Oscillations and Hall Effect in Quasi-2D Graphite Quantum Dots”; J.S. Bunch, Y. Yaish, M. Brink, K. Bolotin, and P.L. McEuen. Nanoletters, 5 (2), 287 (2005).


Figure 1, top right: Optical micrograph of a nanotube device used for high frequency measurements. Nanotubes are grown by chemical vapor deposition from the catalyst pad and source drain and gate electrodes are subsequently patterned on the device. The metal top gate is separated by a 10 nm evaporated silicon oxide dielectric.

Figure 2 below left: Optical micrograph of the “designed electrode” method in which graphite quantum dots or SWNTs are located by AFM with respect to predefined alignment marks etched in the silicon oxide and then contacted by metal leads defined by electron beam lithography.

Figure 3, below right: AFM of a quasi 2D graphite quantum dot connected to multiple metal leads. Electrons in these mesoscopic graphite pieces are delocalized over the whole graphite piece down to the low temperatures.

Carbon NanoElectronicsCNF Project # 900-00 Principal Investigator(s): Paul L. McEuen

User(s):Markus BrinkScott BunchLuke DonevNathaniel GaborLisa LarrimoreSami RosenblattVera SazoanovaArend van der ZandeXinjian ZhouKen BosnickJun ZhuShahal Ilani

Principal Investigator(s):Paul L. McEuen

Affiliation(s):Physics Department, Cornell University

Primary Funding:MARCO/DARPA, CNS, CCMR,NASA, Packard Foundation

Contact Information:[email protected]

http://www.lassp.cornell.edu/lassp_data/mceuen/homepage/welcome.html

Figure 1

Figure 2 Figure 3



Abstract:Due to their unique structure, branched nanostructures such as carbon

nanotube Y-junctions are theoretically estimated to have large piezoresistive coefficients, which in turn could be used for electronic or electromagnetic sensing of nanoscale displacement. Here, we report the fabrication of such a nanoscale sensor using a “divining rod”, branched nanotube structures.

Summary:The Y-junction carbon nanotubes (YCNT) employed have a unique “fish-

bone” structure, with typical diameters of 20 to 50 nm, and length ranging from 1 to 15 µm. The typical resistivity of a straight segment of a fishbone CNT is of the order of 105 Ω/µm, similar to that of the graphite along its c-axis.

To fabricate a nanoscale sensor, we start by making a suspension with YCNTs and dichloride methylene, a small drop of which is placed onto a silicon substrate (the substrate has a 1 µm layer of thermally-grown silicon oxide, prepatterned with micron-size alignment marks for nanofabrication), and carefully allowed to dry in the marks area. SEM is then used to search for a typical YCNT, locating its coordinates relative to the fiducials. Nanofabrication is done with the CNF Leica VB6, forming at least 4 wires across 2 branches of the Y-junction (nanopatterns are designed in L-Edit with coordinates given by SEM).

To make good electrical contacts, we’ve tested a few different metal depositions. We found the best conbination to be 50 nm (or more) Ti + 10nm Au. Here, Ti is the one to contact the YCNTs, Au is used only to prevent Ti from getting oxidized. Other metals, like Pd/Au, may work in some way, but we found Pd/Au didn’t stick to the YCNTs well, as it is so brittle that many times it cracked and shifted away from the nanotubes.

Toward Fabrication of Nanoscale Sensors Using “Divining Rod” Carbon Nanotubes

CNF Project # 1192-04 Principal Investigator(s): Michael J. Naughton

Now, if every step is successful, we should come out with a structure that has at least 4 Ti/Au wires across 2 of the 3 branches of the YCNT. The next step is to suspend the 3rd branch to form a divining rod. This is realized by another round of nanofabrication, followed up by BOE (buffered oxide etching, etchant: 6 parts 40% NH4F and 1 part 49% HF) and a critical point dry.

Finally, we did a rapid thermal anneal [1] at 650˚C for 30 s. With this step, the contact resistance of (Ti & YCNT) drops sharply from ~1 GΩ range to the order of 1kΩ, and becomes very stable.

In conclusion, we have made nanoscale sensors with individual Y-junction CNTs, using photolithography, nanolithography and other techniques. We are now measuring the piezoresistivity of these divining rod YCNTs to low temperatures and in high magnetic fields [2].

References:[1] “Formation of low-resistance ohmic contacts between carbon nanotube and metal electrodes

by a rapid thermal annealing method”, Jeong-O Lee, et al. J. Phys. D: Appl. Phys. 33 (2000) 1953–1956.

[2] 2005 American Physics Society March Meeting talk—P40.00011 “Transport Measurements on Individual Branched Nanostructures”, by Yong Sun.


Figure 1: SEM picture of a Y-junction CNT suspending above a well (~ 400 nm deep, etched by BOE). The nanotube is contacted by 4 metal leads (50 nm Ti + 10 nm Au).

Toward Fabrication of Nanoscale Sensors Using “Divining Rod” Carbon Nanotubes

CNF Project # 1192-04 Principal Investigator(s): Michael J. Naughton

User(s):Yong SunJeong-Il Oh

Principal Investigator(s):Michael J. Naughton

Affiliation(s):Dept of Physics, Boston College

Primary Funding:NSF NIRT Grant # 0210533




Abstract:Various resonant structures combined with lumped and distributed circuit

elements were designed for the purpose of identifying an array configuration to efficiently down-convert frequency of incident radiation both at mm-waves at 100 GHz (λ = 3 mm) and in the long wavelength IR spectrum at around 28.3 THz (λ = 10.6 µm).

This is a first effort for us in fabricating nano-structures, therefore a major part of this effort was aimed at getting acquainted with the tools and methods of nanofabrication. We were looking for answers to questions such as what is realistic and what is not in fabricating thin film circuits with sub 100 nm features? Some other questions were: what are the limits of resolution and limits on circuit complexity today’s e-beam lithograthy technology can support? Of special interest for us was to develop a feel for what are the limits on reproducibility within small areas of a few hundred microns square, and also step and repeat resolution and reproducibility over long distances (over the surface of a 3” or 4” diameter wafer).

Economy (machine use time) for producing large surfaces was another key concern. Answers to all these (and other) questions were very important in the process of determining whether or not any of concepts and novel ideas we have in optical frequency conversion will be practical and realizable.

The patented technique of using connected dipole arrays with diodes and other nonlinear components requires low loss conductive paths at IR frequencies (Au) and low loss, thin dielectric support membranes.

Three different structures were fabricated in two visits to CNF last year. Linear arrays of half-wave dipoles were fabricated in several different configurations using evaporated gold to be used in tests at λ = 10.6 µm. Cold-emission diodes in a thin film strip (Au) configuration were made with various gap dimensions (28 nm and larger). Finally we produced some simple resonant circuits (oval shaped loop antennas) with cold-emission diodes incorporated within.

Development of Resonant Structures at 10 µm WavelengthCNF Project # 1223-04 Principal Investigator(s): Sandor Holly, William D. Mack

Subsequent measurement tests of these structures showed the expected resonances, however these resonances were at lower than calculated values. Dielectric const of support materials had stronger effect than prior estimation.

Summary:Our initial experience at CNF (and with CNF’s very capable staff) answered

many (but not all) of our questions (and concerns) in a positive way. We have become convinced that the presently available nanofabrication technology—as we experienced it at CNF—is able to support the submicroscopic requirements of our needs.

Up till now we have been bonding the additional lumped circuit components we need for full functionality into the microscopic photolithographically made circuits manually. It is an extremely time consuming and difficult undertaking.

In view of the planned future efforts of going to higher frequencies, to shorter wavelength and smaller dimensions still, it is unavoidable to move to total integration where all array components, diodes, transistors etc. are fabricated together just as in any modern integrated circuit manufacturing. This of course will encompass the use of MBE technology at nanometer scales.

References:[1] Holly, S. et. al. Various White Papers, Internal Memos on THz Generation.[2] Codreanu, I. and Boreman, G. “Infrared Microstrip Dipole Antennas” Microwave and

Optical Technology Letters 29, 381-3 (2001).[3] Fumeaux, C., Gritz, M., Codreanu, I., Schaich, W., Gonzales, F., Boreman, G.,

“Measurement of the Resonant Lengths of Infrared Dipole Antennas” Infrared Physics and technology 41, 271-281 (2000).


Figure 1: One dim. array of dipoles• Dipoles lined up along a straight line.

• Length of ea. Dipole = 4.12 µm

• Width of dipole = 170 nm

• Separation between dipoles = 750 nm

• Thickness (Au) = 100 nm

Figure 2: Cold-emission tip• Gold Strip Width = 170 nm

• Gap width = 48 nm

• Tip Radius of Curvature = 12 nm (approx)


Figure 3: Simple Resonant Circuit • Loop major axis = 3.72 µm

• Loop minor axis = 2.15 µm

• Loop conductor width 160 nm

• Gap width at cold emission tip = 48 nm


Development of Resonant Structures at 10 µm WavelengthCNF Project # 1223-04 Principal Investigator(s): Sandor Holly, William D. Mack

User(s):Sandor HollyWilliam D. Mack

Principal Investigator(s):Sandor HollyWilliam D. Mack

Affiliation(s):Laser & Electro-Optical Systems (LEOS), Boeing Corporation

Primary Funding:Boeing IR&D Funds

Contact Information:[email protected]@boeing.com



Abstract:The Cornell Nanoscale Facility (CNF) was used to fabricate arrays of gold

nanowires (33 nm width) attached to macroscopic leads. These wires were then stressed by applying current until they failed under electromigration. With the devices fabricated at CNF, we were able to establish that (a) the electromigration process used to form the nanogap junctions is temperature-controlled, and (b) that good thermal contact of the wire to the substrate or to the electrodes is essential for controlling the electromigration to form uniform size nanogaps [1]. The ultimate goal is to use the CNF structures in future studies of single-molecule conduction across nanogap electrode pairs [2].

Summary:Nanogap electrode structures such as those fabricated at CNF have been

used by the principal investigators to investigate the electronic properties of a novel ferrocene-containing molecular wire [2]. Future investigations of other types of organometallic molecular wires are being planned; the fabrication of these structures en masse at CNF will greatly increase the ability to measure different molecular species rapidly.

In addition to the molecular electronics work, we have also studied the formation process of the nanogap junctions. An electrical feedback technique was developed to controllably break the junctions to form a gap with desired

Nanogap Fabrication for Single-Molecule Electrical MeasurementsCNF Project # 1245-04 Principal Investigator(s): Michael S. Fuhrer, Lawrence R. Sita

(tunnel) resistance (and presumably desired gap spacing). The feedback scheme was found to work only in the case that the nanowire was thermally well-coupled to either the bulk metal reservoirs (i.e. the wire was short) or a gate electrode (in the CNF devices the wire is on top of an aluminum gate electrode with thin alumina dielectric) [1]. It was also found that as the resistance of the junction increases, the junction voltage increases such that R is proportional to V squared; i.e. the power is roughly constant. This implies that the feedback process acts to control the temperature of the wire during electromigration [1].

References:[1] “Formation of Molecular-Scale Gold Nanogap Junctions via Controlled Electromigration,”

G. Esen, S.A. Getty, M.S. Fuhrer, to appear in the Proceedings of the International Winterschool on Novel Properties of Electronic Materials, H. Kuzmany, J. Fink, M. Mehring, and S. Roth, Editors (AIP Conference Proceedings, New York, 2005).

[2] “Near-perfect conduction through a ferrocene-based molecular wire,” Stephanie A. Getty, Chaiwat Engtrakul, Lixin Wang, Rui Liu, San-Huang Ke, Harold U. Baranger, Weitao Yang, Michael. S. Fuhrer, Lawrence R. Sita, Physical Review B Rapid Communications 71, 241401 (2005).


Figure 1:Gold nanowire fabricated at Cornell Nanoscale Facility. This wire has undergone an electromigration process to form a nanogap; the gap is barely visible in this scanning electron micrograph.

Nanogap Fabrication for Single-Molecule Electrical MeasurementsCNF Project # 1245-04 Principal Investigator(s): Michael S. Fuhrer, Lawrence R. Sita

User(s):Stephanie A. Getty

Principal Investigator(s):Michael S. FuhrerLawrence R. Sita

Affiliation(s):Physics and Chemistry, and Center for Superconductivity Research, University of Maryland, College Park, MD

Primary Funding:NSF-NIRT


h t t p : / /www.phys i c s . umd .edu /mfuhrer/



Abstract:We report the morphology of gold clusters in contact with carbon nanotubes

using electron tomography which provides three-dimensional information at the spatial resolution of 2 nm3. Investigating the three-dimensional morphology of and the interfacial structure between gold clusters and carbon nanotubes is important as they influence the contact potential significantly. From the tomographically reconstructed three-dimensional data, we observe that the gold clusters facet and the interfacial plane between the gold clusters and the carbon nanotubes is flat. The flat interfacial plane leads to deformation of the carbon nanotubes at regions in contact with the gold clusters. We also report that nucleation energy of the gold clusters is inversely proportional to the diameter of the nanotubes. Furthermore, the gold clusters elongate along the long axis of the carbon nanotubes as the surface across the nanotubes is highly curved to increase the nucleation energy of the gold clusters.

Experimental Details:Single-walled nanotubes were dispersed in pure hexane solvent and

sonicated to isolate the individual nanotubes. A drop of the solution was applied to a TEM grid that is coated with a thin holey carbon film. Then four monolayers of gold were deposited on the nanotubes by E-beam evaporation. A scanning electron transmission microscope (STEM) which provides an atomic resolution of 0.16 nm was used to investigate the structure of the gold-deposited carbon nanotubes. The gold-deposited carbon nanotubes were tilted from -75° to + 75° at a tilt increment of 2° and a projected image at each tilt angle was acquired to generate a tilt series [1]. The acquired tilt series was then used to reconstruct a three-dimensional representation of the gold-deposited nanotubes. The morphology of the gold clusters[2, 3] reconstructed from the tilt series was investigated as a function of the diameter of the carbon nanotubes [4] as well as a function of the number of monolayers of gold deposited on the nanotubes. Nanodiffraction patterns and high resolution TEM images of the gold-deposited nanotubes were taken as well to complement the tomographically reconstructed data.

Contact Geometry of Gold Clusters on Carbon Nanotubes using Electron Tomography

CNF Project # 1246-04 Principal Investigator(s): David A. Muller

Different metallic elements such as gold-palladium alloy, nickel, iridium, and etc., will be deposited on the carbon nanotubes [5] to observe different morphologies of these ad-clusters from which the contact potential can be deduced.

Conclusions: Investigating the three dimensional morphology of an interfacial structure

between the gold clusters and the carbon nanotubes using electron tomography has been successfully achieved. The experimental results show faceted gold clusters and flat interfacial plane between the gold clusters and the carbon nanotubes. The flat interfacial plane signifies deformation of the carbon nanotubes at regions in contact with the gold clusters. Moreover, the nucleation energy of the gold clusters increases with decrease in the diameter of the carbon nanotubes. From the structural investigation on different metallic adatoms on carbon nanotubes with electron tomography, we hope to develop a method to minimize the contact potential between metallic leads and carbon nanotubes.

References:[1] P.A. Midgley, M. Weyland, ‘3D electron microscopy in the physical sciences: the

development of Z-contrast and EFTEM tomography’, Ultramicroscopy, 96: 413-431 (2003).

[2] N. Doraiswamy, L. D. Marks, ‘Preferred structures in small particles’, Phil. Mag. B, 71 : 291-310 (1995).

[3] C. Herring, ‘Some theorems on the free energies of crystal surfaces’, Phys. Rev. 82: 87-93 (1951).

[4] O. Gülseren, T. Yildirim, S. Ciraci, ‘Tunable Adsorption on Carbon Nanotubes’, Phys. Rev. Lett., 87: 116802 (2001).

[5] Y. Zhang, H. Dai, ‘Formation of metal nanowires on suspended single-walled carbon nanotubes’, Appl. Phys. Lett., 77: 3015-3017 (2000).


• Gold clusters facet on carbon nanotubes.

• Interfacial plane between gold clusters and nanotubes is flat.

• Carbon nanotubes deforms due to the flat interfacial plane.

Contact Geometry of Gold Clusters on Carbon Nanotubes using Electron Tomography

CNF Project # 1246-04 Principal Investigator(s): David A. Muller

User(s):Jeeyoung Cha

Principal Investigator(s):David A. Muller

Affiliation(s):Applied & Engineering Physics, Cornell University

Primary Funding:CCMR, IRG-A


http:/ /people.ccmr.cornell .edu/~davidm/

Figure 1: Gold clusters on carbon nanotubes of different diameters (STEM-HAADF images) show that the gold clusters elongate along the nanotubes.

Figure 3: Tomographically reconstructed gold clusters on carbon nanotubes reveal that the interfacial plane between the gold clusters & nanotubes is flat.

Figure 2: Hi-res TEM image shows that the gold cluster on the carbon nanotube is strongly faceted.

nanosecond switching behavior of nanopillar spin valves · nanosecond switching behavior of...

Documents